Re: [PATCH 1/2] vfio/container: Support unmap all in one ioctl()

["Cédric Le Goater" <clg@redhat.com>]

On 10/1/25 00:04, John Levon wrote:
> On Tue, Sep 30, 2025 at 05:26:59PM +0200, Cédric Le Goater wrote:
> 
>>> -        }
>>> +        ret = ioctl(container->fd, VFIO_CHECK_EXTENSION, VFIO_UNMAP_ALL);
>>
>> Could we introduce an helper to test 'unmap_all' support in the host
>> kernel ? The result would be something like :
>>
>>    if (unmap_all) {
>>          if (vfio_legacy_has_unmap_all(VFIO_IOMMU_LEGACY(bcontainer))) {
>>              flags = VFIO_DMA_UNMAP_FLAG_ALL;
>>          } else {
>>              /* The unmap ioctl doesn't accept a full 64-bit span. */
>>              Int128 llsize = int128_rshift(int128_2_64(), 1);
>>              ...
>>          }
>>    }
> 
> For reference/consideration, the previous approach taken in the vfio-user
> series:
> 
> https://lore.kernel.org/qemu-devel/20250219144858.266455-4-john.levon@nutanix.com/
> 
> @@ -533,6 +562,11 @@ static bool vfio_legacy_setup(VFIOContainerBase *bcontainer, Error **errp)
>       vfio_get_info_iova_range(info, bcontainer);
>   
>       vfio_get_iommu_info_migration(container, info);
> +
> +    ret = ioctl(container->fd, VFIO_CHECK_EXTENSION, VFIO_UNMAP_ALL);
> +
> +    container->unmap_all_supported = (ret != 0);
> 
> (I dropped this particular change as part of getting merged.)
Yes. I’m reconsidering now.

Should we introduce a VFIOContainerBase attribute/flag 'unmap_all_supported',
set in the vioc->setup handler ?

Thanks,

C.